Re: Unit Testing & Test-Driven Development (was Re: GSWeb state?)

[Richard Frith-Macdonald]

On Saturday, November 30, 2002, at 06:45  pm, Richard Frith-Macdonald 
wrote:

> On Saturday, November 30, 2002, at 03:30  am, Chris Hanson wrote:
>
> > At 9:38 PM +0100 11/27/02, Manuel Guesdon wrote:
> > >         o It as has a lot of features so it's not completely tested 
> > > (we had
> > > recently an exemple of different  people using GSWCheckBoxList with
> > > different bindings. Someone has fixed what it  considered as a bug so
> > > it worked for him but no more for me so I had to rewrite some part 
> > > of code to be
> > > sure that it works for both of us)
> >
> > Are the developers working on the various pieces of GNUstep creating 
> > comprehensive unit tests for the code they're writing?  (I.e. by 
> > using a unit-testing framework like OCUnit from Sen:te and having all 
> > tests run every build?)
>
> This is only in place for the base library really ... and can hardly 
> be called 'comprehensive' as there are only a few thousand tests.
> See dev-apps/test/gtests in CVS
>
> People are of course encouraged to contribute tests, but it's no more 
> glamorous than writing documentation ...

PS. The tests for the base library use a posix compliant regression 
testing package I wrote for the purpose called 'greg'
This uses guile as a scripting language to execute objc methods 
directly so it's somewhat more dynamic and easier to write and group 
testcases than in OCUnit, but the basic functionality is pretty similar.